A Tale of Three Cities: Digital Redlining as it Pertains to Access and Quality

The research regarding equality of internet access has been split into two separate, but related, concepts: the American digital divide, and digital redlining. The American digital divide primarily pertains to the inequalities present between groups of people in terms of access to the internet, and knowledge and skill needed to interact with the internet ¹. Although research shows that the digital divide has reduced considerably as more Americans develop digital literacy skills, there are still considerable gaps when grouping based on demographic characteristics ².

Digital redlining, on the other hand, “...[I]s the creation and maintenance of technology practices that further entrench discriminatory practices against already marginalized groups” ³. The nuance is quite specific- this isn’t simply about the difference between accessing and knowing how to access the internet, but an exploration of the systemic inequalities that are baked into internet access that, then, result in dissimilar levels of access based on race. Digital redlining is the maintenance and expansion of the American digital divide, as well as the admission that the internet is systemic in its inequality of access in America ⁴.

This practice is predominately apparent within the context of the inner city neighborhood although the way it actualizes differs from city to city. It could either be through ISPs not offering high speed internet packages to areas with high poverty rates like in Cleveland and Dallas. Or it could be by the uneven level of distribution where ISPs are disproportionately focusing on providing services in high income, predominately white, neighborhoods. This latter form of digital redlining has been found in Baltimore, Cleveland , Detroit, Los Angeles, Fresno, Oakland, and Dallas to name a few.

Although the geographic impacts of “classic redlining” (withholding mortgage credit) has been explored through a variety of lenses, this research question has not been applied to digital redlining. That is to say, how did the Home Owners’ Loan Corporation (HOLC) neighborhood rating map impact both the accessibility and the quality of internet service? Of those living in HOLC neighborhoods, does the HOLC grade impact internet accessbility or internet quality? What does accessbility and quality look like for different racial and/or ethnic groups in different HOLC neighborhoods?

In order to explore these questions we will be comparing digital redlining across three different cities: Baltimore, Detroit, and Fresno.

There is a fantastic project called Not Even Past: Social Vulnerability and the Legacy of Redlining where researchers from the University of Richmond, mapped out the redlining maps made by the HOLC during the 1930s, onto the present day census tracts. This particular dataset also gives us a percentage as to how much of a present day census tract falls under the HOLC neighborhood map as well. Utilizing this variable we can make an estimate as to how many people currently living in a HOLC neighbhorhood are being impacted by digital redlining.

When we graph this question of internet accessibility by race and/or ethnic group, HOLC grade, and city on the same scale, there are some immediate, and obvious takeaways.

Although the three cities are of roughly the same size (500000 – 600000 people), Baltimore and Detroit have far more people living in HOLC rated neighborhoods that do not have internet accessibility. We can also see this in the interactive maps below where, in Baltimore and Detroit, the HOLC redlining maps cover the entire expanse of the city; in Fresno the HOLC maps only covers a portion of the entire city. If we change the scale we can see how digital redlining impacts the different HOLC residents in these cities.

In Fresno, the racial and/or ethnic group with the largest population that does not have internet accessibility is the Hispanic and/or Latinx people. This is an unsurprising result when you consider the fact that Hispanic and/or Latinx people make up 49.6% of the population. What is more surprising, and a theme we will see throughout, is how the white population (despite being 60% of the overall population in Fresno) has a much lower number of people without internet accessibility in HOLC neighborhoods. Of the Hispanic and/or Latinx people who don't have internet accessibility, they largely live in neighborhoods that were formerly labelled as “Declining” (grade C) or redlined (grade D). In Detroit it is the Black HOLC residents that do not have internet accessibility.

Interestingly enough, in Baltimore, the HOLC score does not matter as Black residents living in “Still Desirable” (grade B) neighborhoods are also impacted by lack of internet accessibility. What these graphs show us is that the HOLC score does appear to impact digital redlining as well. There are large populations of Black, and Hispanic and/or Latinx people who live in “Declining” (grade C) or previously redlined (grade D) neighborhoods, and do not have internet accessibility.

While poverty might be an explanation for why grade C and grade D neighborhoods appear to have the largest number of people with no internet, we would also need to see a larger proportion of white residents being affected in these areas as well. For such a large group of people, the white population of these cities do not appear to be as affected by digital redlining as the Black and Hispanic and/or Latinx populations. Although the numbers of no availability are higher for white residents in grade C and grade D neighborhoods, it is not nearly at the same scale as the Black and Hispanic and/or Latinx population.

As mentioned earlier, digital redlining is not only about internet accessibility but also about the quality of the internet that is being offered.

As we can see in this graph above, there is a negative average trend of median download speeds by HOLC scores. Of those living in HOLC neighborhoods, those in Grade D neighborhoods have the worst median download speeds.

Taken all together, we have three different stories of digital redlining and its manifestations. In Baltimore, we can see from the static visualization that Black residents seem to be disproportionately affected by internet accessibility regardless of where they live. Not only does Detroit have the largest population of residents who have no internet (Black residents), but it also the city with the lowest median speed. In Fresno, it’s the Hispanic and/or Latinx population that mostly does not have internet accessibility. Not only that but there is also a dramatic difference in median speeds by HOLC neighborhoods. And, across all three cities, neighborhoods that had a grade of C or D have the highest number of residents without internet access.

But--what does this look like in real life? When we say there are populations that do not have internet in an area-- what does this look like? In order to contextualize the above information I've created 3 digital redlining maps, below, which outputs the HOLC grading on top of modern day census tracts. A codebook for the variables included is provided below.